Modeling Intraday Momentum and Reversal by a Parsimonious Diffusion Process

Abstract

We extend the double-well potential process to a three-parameter version in order to model intraday price dynamics, with a focus on the intraday momentum and reversal. The proposed process has a parsimonious form of three parameters controlling momentum, reversal, and volatility respectively. By using different combinations of parameter values, the process can accommodate two different intraday patterns: i) there is no new information and intraday price undertakes mean-reverting pattern due to “noise traders”; ii) there is new information released and the intraday price experiences a momentum possibly followed by a reversal due to overreaction. Estimation methods are developed under both long-span asymptotics and in-fill asymptotics along with a simulation study for the finite sample properties. An empirical illustration is provided by using three intraday-frequencies data of Apple Inc.